Channel Estimation using Approximate Conjugate Gradient Pursuit for Hybrid MIMO System in Millimeter Wave Communication

Millimeter wave (mmWave) based systems can achieve very high data rates, thanks to large bandwidth and large antenna array systems which make it possible at such high frequencies. But at these high frequencies, analog front end devices are very much power hungry. Due to cost and power constraints, hybrid architecture needs to be deployed for mm precoding. mm Wave channel specific estimation techniques are required to estimate the channel due to the presence of a large number of antennas and analog beamformers. In this paper, the parametric channel estimation problem is casted into a sparse recovery problem considering the sparsity of mm Wave channels and is solved using approximate conjugate gradient pursuit (ACGP). The proposed method is fast and less complex as compared to the orthogonal matching pursuit (OMP) used in the literature. Performance comparison has been done in terms of floating point operations and time required by both algorithms for calculating the channel. The proposed method takes O(mn) computations as opposed to OMP which takes O(n3) computations (n > > m), thereby reducing a large number of computations especially for a system having a large array of antennas as used in present day technology.